From owner-freebsd-stable@freebsd.org  Thu Jul 19 09:45:06 2018
Return-Path: <owner-freebsd-stable@freebsd.org>
Delivered-To: freebsd-stable@mailman.ysv.freebsd.org
Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2610:1c1:1:606c::19:1])
 by mailman.ysv.freebsd.org (Postfix) with ESMTP id 60E17102AAFF
 for <freebsd-stable@mailman.ysv.freebsd.org>;
 Thu, 19 Jul 2018 09:45:06 +0000 (UTC)
 (envelope-from petefrench@ingresso.co.uk)
Received: from constantine.ingresso.co.uk (unknown [IPv6:2a02:b90:3002:411::3])
 (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
 (Client did not present a certificate)
 by mx1.freebsd.org (Postfix) with ESMTPS id F1E1B88229
 for <freebsd-stable@freebsd.org>; Thu, 19 Jul 2018 09:45:05 +0000 (UTC)
 (envelope-from petefrench@ingresso.co.uk)
Received: from [2a02:b90:3002:411::6] (helo=dilbert.ingresso.co.uk)
 by constantine.ingresso.co.uk with esmtpsa
 (TLSv1.2:ECDHE-RSA-AES256-GCM-SHA384:256) (Exim 4.89 (FreeBSD))
 (envelope-from <petefrench@ingresso.co.uk>) id 1fg5Ue-000Erp-5X
 for freebsd-stable@freebsd.org; Thu, 19 Jul 2018 09:45:04 +0000
Received: from petefrench by dilbert.ingresso.co.uk with local (Exim 4.91
 (FreeBSD)) (envelope-from <petefrench@ingresso.co.uk>)
 id 1fg5Ue-0009ot-42
 for freebsd-stable@freebsd.org; Thu, 19 Jul 2018 10:45:04 +0100
To: freebsd-stable@freebsd.org
Subject: An oddity of memory speeds and timings
Message-Id: <E1fg5Ue-0009ot-42@dilbert.ingresso.co.uk>
From: Pete French <petefrench@ingresso.co.uk>
Date: Thu, 19 Jul 2018 10:45:04 +0100
X-BeenThere: freebsd-stable@freebsd.org
X-Mailman-Version: 2.1.27
Precedence: list
List-Id: Production branch of FreeBSD source code <freebsd-stable.freebsd.org>
List-Unsubscribe: <https://lists.freebsd.org/mailman/options/freebsd-stable>, 
 <mailto:freebsd-stable-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-stable/>
List-Post: <mailto:freebsd-stable@freebsd.org>
List-Help: <mailto:freebsd-stable-request@freebsd.org?subject=help>
List-Subscribe: <https://lists.freebsd.org/mailman/listinfo/freebsd-stable>,
 <mailto:freebsd-stable-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Thu, 19 Jul 2018 09:45:06 -0000

Yesterday I inreased my memory speed on my Ryzen box from 2133 to 2400
as I had previously been underclocking it. Ryzens are sensetive to
memory clock speed as it affects the speed of the underlying fabric
between the cores as well as I understand it, so worth running it at
its rated speed.

I meausred the improvement by using 'time' on a CPU bound compile. It
speeded wall clock time up by abotu the expected amount, but what I
found curious looking at the breakdown of the timings reported by
time was that. there was no reduction in user time, but a drastic
reduction (25%!) in system time. Can anyone explain that ?

My only wild theory at the moment is that for a single process the user space
component is single threaded and shows no real improvement, but the
kerenl, being multi-threaded, will benifit from the speedup of the
interconnect between the cores. But I dont know if the timings measures by
time wuld actually show that.

Interesting though.

-pete.